Zipf distribution

#1
I have an unknown zipf distribution with unknown number of objects. Suppose I select n1, n2…samples from the distribution and I get u1, u2…unique samples respectively, can I determine the parameters of the distribution including the total number of objects? Is it possible to obtain the confidence intervals?
 

katxt

Active Member
#3
Does your data look like this? ...
3rd most likely 270 things
5th most likely 60 things
10th most likely 8 things
 
#4
Thanks hlsmith and katxt for the reply.
Yes, I can dertermine nth most likely m objects from the data I have. Will n=N and m=k in zipf's distribution? How do I determine the exponent 's'?